Вы находитесь на странице: 1из 10

245

Design of a lightweight automotive brake disc using


finite element and Taguchi techniques
D G Grieve1 , D C Barton1 , D A Crolla1 and J T Buckingham2
Department of Mechanical Engineering, University of Leeds
2
Rover Cars Limited, Leeds

Abstract: Aluminium metal matrix composite brake discs offer significant weight advantages compared
with the traditional cast iron rotor but have a much lower maximum operating temperature. In this study, a
finite element model of an existing brake design is firstly used to predict the peak disc temperatures during
two critical brake tests for both cast iron and an aluminium metal matrix composite alternative. A Taguchi
analysis is then applied, enabling all the critical design and material factors of an aluminium metal matrix
composite rotor to be considered collectively. Based on the results of this exercise, a parametric sensitivity
study is carried out to define suitable designmaterial combinations for a prototype lightweight front brake
disc to be used on small to medium passenger vehicles.
Keywords: lightweight automotive brake disc, finite element model, Taguchi analysis, parametric
sensitivity study
NOTATION
A
d
Cp
Fs
g
h
hd
k
l
p
qc
qr
_
Q
m
Nu
r
Re
s
t
vf
vi
V

area of one rubbing surface (m2 )


drag losses (per cent)
specific heat (J=kg K)
velocity reduction factor
gravity (m=s2 )
heat transfer coefficient (W=K)
height dropped during descent (m)
conductivity (W=m K)
characteristic surface length (m)
heat partitioned to the pads (%)
convective heat flux (J=s)
radiative heat flux (J=s)
total heat flux (J=s)
mass of the vehicle (kg)
Nusselt number
radius of the brake disc (m)
Reynolds number
brake split to front discs (%)
time(s)
final speed of the vehicle (m=s)
initial speed of the vehicle (m=s)
free-stream velocity (m=s)

StefanBoltzmann constant
emissivity

The MS was received on 19 March 1997 and was accepted for publication
on 17 November 1997.

r
1

temperature (8C)
surrounding fluid temperature (8C)
kinematic viscosity of air (m2 =s)
density (kg=m3 )
INTRODUCTION

Legislation due to be introduced by the year 2010 will


require vehicle emissions of HC, CO2 and NO x to be
reduced to 60 per cent of 1996 levels. This has fuelled
research into the reduction of vehicle masses by the use of
lightweight components. In particular, the replacement of
the conventional cast iron brake rotor with an aluminium
metal matrix composite (Al-MMC) alternative has recently
received attention. Initial work has outlined the problems of
overheating and softening of existing Al-MMC materials
due to their relatively low melting points [14]. However,
the launch of the Lotus Elise, heralded as the first
passenger vehicle to utilize Al-MMC materials for both the
front and the rear brake rotors, clearly shows the potential
of this technology. The all-aluminium space frame design,
together with restricted passenger and luggage space all
help to contribute to the Elise's exceptionally low gross
vehicle mass (GVM) of only 650 kg. This has enabled
Lotus to specify a 30 vol % SiC-particulate-reinforced AlMMC rotor which operates well below the material's
critical temperature on both the front and the rear brake
assemblies but still gives a saving in mass of about 40 per
cent over conventional cast iron rotors.

D01097 # IMechE 1998

Proc Instn Mech Engrs Vol 212 Part D


Downloaded from pid.sagepub.com at CIDADE UNIVERSITARIA on March 27, 2015

246

D G GRIEVE, D C BARTON, D A CROLLA AND J T BUCKINGHAM

The challenge now lies in applying this technology to


high-volume-production smallmedium family saloons
which have smaller wheels and much higher GVMs than
the Elise does. The front discs on a typical European
family saloon have a mass of approximately 5.25 kg each.
Since the density of most Al-MMCs is about one third
that of cast iron, such materials have the potential (if
used on both the front and the rear brakes) to save
approximately 12 kg per vehicle. As well as saving mass,
Al-MMC brake discs may offer refinement advantages
such as reduction in squeal, judder and wear, leading to
prolonged rotor life.
In previous work [5, 6], a thermal finite element
analysis (FEA) has been shown to be an efficient and
accurate method of estimating the peak disc temperatures
during critical vehicle brake tests. In the present paper,
these techniques are used to investigate the design
modifications and material specifications to enable the
use of Al-MMC brake discs on the front of a typical
smallmedium passenger car. Three separate studies are
described. The first examines the effect of the vehicle
mass on the peak disc temperatures for two onerous but
quite different critical brake tests: a prolonged Alpine
descent and a rapid high-speed autobahn stop. This is
followed by a more comprehensive study using the
Taguchi technique to find the most influential factors to
be considered for the design of a prototype Al-MMC
brake disc. Finally a parametric study is conducted in
which the effect of disc cheek thickness and vent width
is examined in detail.

Fig. 1

FINITE ELEMENT MODELS

All finite element (FE) models described below were


analysed using the ABAQUS software package running
on a Sun SPARC 10 model 51 workstation. A threedimensional FE model of a 108 segment of a vented
front disc, hub and wheel assembly for a typical
medium-sized passenger car was generated using quadratic heat transfer brick elements so as to include one
vent and two half-vanes (Fig. 1), giving a total of 1134
elements with 18 824 nodes giving 18 824 degrees of
freedom. The disc wheel and hub were combined into
one solid model so as to enable conductive heat transfer
to be carried out between the disc, wheel and hub.
Sensitivity studies to investigate the thermal resistance
between the hub and disc and the wheel and disc
surfaces revealed that little or no thermal resistance was
apparent (owing to mechanical clamping of the wheel to
the hub and disc). The wheel was truncated in order to
reduce the complexity of the three-dimensional model
after sensitivity studies using two-dimensional models
revealed that little heat is conducted to the outer areas
of the wheel. The inclusion of the hub was necessary so
as to enable estimation of hub temperatures which may
cause the bearing lubricant to break down. The truncated
end of the wheel, the internal diameter of the hub, and
the under surfaces of the hub were treated as free
surfaces so as to model the small amount of heat that is
conducted to and subsequently lost from these parts in
the geometry.

FE mesh of a 108 segment of a vented brake disc, hub and truncated wheel

Proc Instn Mech Engrs Vol 212 Part D

D01097 # IMechE 1998


Downloaded from pid.sagepub.com at CIDADE UNIVERSITARIA on March 27, 2015

DESIGN OF A LIGHTWEIGHT AUTOMOTIVE BRAKE DISC USING FE AND TAGUCHI TECHNIQUES

Convective and radiative heat transfer was applied to all


the free surfaces of the model as follows. The convective
heat flux, qc , from the model's free surfaces is given by
qc h(

0 )

(1)

The convective heat transfer coefficient, h, was found from


the Nusselt modulus, Nu. Vehicles travelling at speeds
above 20 mile=h are thought to give rise to turbulent air
flow at the disc surfaces since the Reynolds number, Re,
will exceed 250 000 where a transition from laminar to
turbulent flow will take place [7]. For a rotating disc in a
cross-flow under turbulent conditions, the Nusselt number
is given by [7]
Nu

hr
0:037Re 0:8
k

(2)

The Reynolds number during forced turbulent conditions


can be found from
Re

2VFs l

(3)

The kinematic viscosity, , and the conductivity of air, k,


were calculated from the average of the ambient air
temperature, 0 , and the disc surface temperature, . The
characteristic surface length, l, was assumed to be the
radius of the brake disc, r, and the free-stream velocity, V,
was assumed to be the speed of the moving vehicle.
Convective heat transfer coefficients were calculated for
all free surfaces of the geometry using equations (2) and
(3). Adjustment of the free-stream velocity, V (and hence
the resulting heat transfer coefficient), by the use of a
velocity reduction factor, Fs , was necessary to take into
account the shielding of the interior vane surfaces, hub and
under wheel surfaces by other vehicle components. Values
of Fs between 0.2 and 0.5 were applied selectively to
different areas of the FE model's surfaces. An empirical
approach was used to determine these values by comparing
cooling coefficients obtained from FEA simulations of a
simple cooling test with measured values [5].
Radiative heat transfer increases with the fourth power
of the surface temperature as shown in the equation below.
Black-body radiation was assumed and the radiative heat
flux, qr , calculated as shown:
qr (4

40 )

(4)

where is the StefanBoltzmann constant.


From thermal imaging work carried out on a brake
dynamometer, the emissivity, , was found to be approximately 0.4 for a cast iron disc with a dark transfer film
[8]. This value of emissivity was used to calculate
the radiative heat transfer from most of the free surfaces
of the model for FEA simulations having both cast iron
and MMC discs. Radiative heat transfer was not used on

247

the vane interior surfaces because of their enclosed


nature.
2.1

Modelling the Alpine descent

_ for the Alpine descent was calculated


The heat flux, Q,
using the equation shown below:
_ mghd s(1
Q

p)(1
4At

d)

(5)

where m is the mass of the vehicle, hd the height dropped


during the descent, t the time taken to descend the
mountain and A the area of one rubbing surface.
Heat was applied as a heat flux to the rubbing surfaces of
the disc, assuming that only 5 per cent of the heat, p,
generated is transferred to the pad. Drag losses, d, for a
vehicle travelling in fourth gear were found to be about
27.4 per cent. The brake split, s, for the vehicle was 72.5
per cent to the front discs. The accuracy of these
assumptions has been demonstrated in previous work
[5, 6].
2.2

Modelling high-speed stops (autobahn stop)

_ for the autobahn stop was calculated at


The heat flux, Q,
0.5 s intervals over the 7.3 s stop from the following
equation:
2
_ m(v i
Q

v2f )s(1 p)(1


4At

d)

(6)

where vi and vf are the vehicle's initial velocity and final


velocity respectively.
The heat transfer coefficients were calculated from
equations (2) and (3) which allow for the constantly
changing vehicle speed and disc surface temperatures.

EFFECT OF THE VEHICLE MASS UPON THE


PEAK BRAKING TEMPERATURES

In this study, FEA simulations of the Alpine descent and


autobahn stop were used to gauge the effect of the vehicle
mass upon the peak temperatures for the front brake of a
medium-sized passenger vehicle. In addition to the standard cast iron, a generic Al-MMC disc material with
20 vol % SiC reinforcement in a typical aluminium
casting alloy matrix (10% Si) was considered with a
density of 2800 kg=m3 , a specific heat of 800 J=kg K and
a conductivity of 180 W=m K. The Al-MMC's maximum
operating temperature (MOT) was assumed to be 450 8C.
The MOT of a brake rotor material is defined as the
temperature above which severe gouging and galling or
shear of the friction surface occurs owing to the action of
the pad.

D01097 # IMechE 1998

Proc Instn Mech Engrs Vol 212 Part D


Downloaded from pid.sagepub.com at CIDADE UNIVERSITARIA on March 27, 2015

248

D G GRIEVE, D C BARTON, D A CROLLA AND J T BUCKINGHAM

Reduction in the GVM was predicted to cause Alpine


descent temperatures to fall significantly for both the cast
iron and the Al-MMC discs (Fig. 2). The lower temperatures for the latter are due to the superior conductivity of
the disc material which over this prolonged test (23 min)
conducted heat away from the disc cheeks to other parts of
the disc, wheel and hub assembly. The results show that the
temperatures in the bell, hub and wheel bolt-up areas of the
Al-MMC model were higher than those seen in the cast
iron. Heat from these areas was eventually lost to the
atmosphere by convection and radiative heat transfer. The
results indicate that the vehicle mass must be less than
1550 kg in order to prevent the Al-MMC from exceeding
its MOT.
The autobahn stop results (Fig. 3) show that Al-MMC
discs will reach higher temperatures than their cast iron

counterparts. This is due to the short time duration of this


stop (7.23 s) during which all the thermal energy is stored
in the disc. The maximum temperature is therefore
dependent upon the thermal capacity of the disc material.
The lower thermal capacity of the Al-MMC disc (rC p
for Al-MMC 2240 kJ=m3 K; rC p for cast iron 3132
kJ=m3 K) explains the higher temperatures seen. A vehicle
having a mass less than 1100 kg would allow an Al-MMC
disc to survive this test. It must be noted that this assumes
no optimization of the disc geometry in order to take
advantage of the Al-MMC material properties. The temperatures could be further reduced by choosing a more
suitable Al-MMC (the generic material properties were on
the conservative side), by improving the disc design (e.g.
thicker cheeks) or by providing extra cooling to the disc
surfaces.

Fig. 2 Effect of the vehicle mass upon the peak Alpine descent temperature

Fig. 3 Effect of the vehicle mass upon the maximum autobahn temperature
Proc Instn Mech Engrs Vol 212 Part D

D01097 # IMechE 1998


Downloaded from pid.sagepub.com at CIDADE UNIVERSITARIA on March 27, 2015

DESIGN OF A LIGHTWEIGHT AUTOMOTIVE BRAKE DISC USING FE AND TAGUCHI TECHNIQUES

TAGUCHI STUDY

The initial study reported above indicates that the concept


of using Al-MMC discs is feasible for small and low-mass
passenger vehicles. By making various changes to the
brake design and by optimizing the disc material, the
critical temperatures might be further reduced, thereby
allowing for a greater factor of safety and/or allowing
medium-sized passenger vehicles to be accommodated.
However, the number of factors that would need to be
considered collectively to enable a valid study would
require a large number of numerical experiments. In order
to simplify this exercise, a Taguchi technique was
employed to find the design and material factors which
most greatly influence the brake disc's thermal performance. This not only reduces the number of simulations
necessary to optimize the performance but also produces a
robust design by finding the factors which are not greatly
influenced by external or uncontrollable factors.
Taguchi [9] defined factors as either internal or
external. External factors are those which cannot be
designed into a product. Internal factors are those which
can. From previous sensitivity studies [5, 6], six internal
factors were identified to be important in determining the
disc mass and the peak temperature (Table 1). The
maximum and minimum values were chosen for each of
the factors based upon what were thought to be reasonably
achievable. A medium-sized vented front brake disc
geometry was again chosen for this study with a GVM of
1830 kg.
The main external factor for a brake disc is the braking

Table 1

Factors selected for the Taguchi study with their maximum and minimum levels
Levels
Factor

A
B
C
D
E
F

1 (minimum) 2 (maximum)

Cheek thickness (mm)


Vent width (mm)
Density (kg=m3 )
Conductivity (W=m K)
Specific heat (J=kg K)
Cooling rate (factor of present rate)

7.0
1.0
2700
90.0
800.0
1.0

12.0
11.0
3100
240.0
900.0
2.0

249

duty that it will be subjected to. It was therefore decided to


simulate brake tests of a demanding nature which represent
both ends of the braking spectrum:
(a) a long continuous drag stop, i.e. the Alpine descent
lasting 23 min;
(b) a high rate of thermal input stop, i.e. the autobahn stop
lasting 7.3 s.
The L8 Taguchi matrix was chosen to `define' the
numerical experiments (Table 2), since this allows each of
the six factors to be examined at their highest and lowest
values. Two sets of simulations using the Alpine descent
and autobahn stop to represent both extremes of braking
duty were carried out. Each set consisted of eight runs with
the six factors (the factor G is redundant in this example)
set to level 1 or 2 (maximum or minimum) as laid down by
the L8 matrix. The values used for levels 1 and 2 are given
in Table 1. The L8 matrix does not allow for interactions
between the factors. A more complex L27 matrix with
enough columns to allow interactions between the six
factors to be examined is available but would require a
much larger number (54) of numerical experiments to be
carried out.
The predicted maximum temperatures for the 16 simulations to satisfy the reduced L8 Taguchi analysis are shown
in Table 3. The means, variances and signal to noise ratios,
S/N, were calculated for each run; the S/N ratios were
calculated from the following equation based on the
`smaller is better' algorithm [9]:
S

10 log10 V

(7)

in which the variance, V, is given by


V

1 2
(Y Y 22 Y 23    YN2 )
N 1

(8)

where N is the number of noise variables and Yi is the


result from each test (i.e. the maximum disc temperature).
The S/N ratio is a good measure of performance since it
is sensitive to both the mean and the variance. The
`responses' of the S/N ratios were found by finding the
means of the S/N ratios given in Table 3 for each level of

Table 2 L8 Taguchi matrix


Level for the following factors
Run

A,
cheek thickness

B,
vent width

C,
density

D,
conductivity

E,
specific heat

F,
cooling rate

G,

1
2
3
4
5
6
7
8

1
1
1
1
2
2
2
2

1
1
2
2
1
1
2
2

1
1
2
2
2
2
1
1

1
2
1
2
1
2
1
2

1
2
1
2
2
1
2
1

1
2
2
1
1
2
2
1

1
2
2
1
2
1
1
2

D01097 # IMechE 1998

Proc Instn Mech Engrs Vol 212 Part D


Downloaded from pid.sagepub.com at CIDADE UNIVERSITARIA on March 27, 2015

75.24
75.24
67.24
67.24
51.62
51.62
54.10
54.10
57.2
55.1
54.9
54.5
53.9
52.0
52.5
53.3
528 129
322 571
309 193
280 798
243 660
156 822
177 491
212 153
719
547
573
529
489
395
421
461
826.4
699.2
655.7
566.4
420.4
425.7
442.2
459.9
611.0
395.3
434.1
490.7
557.3
363.9
399.3
461.3
800
900
800
900
900
800
900
800
90
240
90
240
90
240
90
240
2700
2700
3100
3100
3100
3100
2700
2700

1.0
2.0
2.0
1.0
1.0
2.0
2.0
1.0

Autobahn,
maximum
temperature
(8C)
Alpine,
maximum
temperature
(8C)
F,
cooling
rate
E,
specific
heat
(J=kg K)
D,
conductivity
(W=m K)
C,
density
(kg=m3 )

each factor examined (e.g. the response for factor B, level


1, was found by finding the mean S/N from rows 1, 2, 5
and 6). The `main effects' were then calculated by finding
the difference between the responses for levels 1 and 2. The
higher the main effect, the greater is the influence that the
factor has upon the disc brake temperatures. Table 4 shows
the responses and main effects for each factor. It can be
seen that factors A, D and F have the strongest main effects
(values shown in parentheses) in the following order of
importance:
(a) cheek thickness (2.5),
(b) cooling rate (1.1),
(c) conductivity (0.9).
The maximization of each of these factors is therefore
recommended in order to minimize peak braking temperatures.
Examining Table 3 it can be seen that run 6 gave the
lowest mean peak temperatures together with the lowest
variance and S/N ratio. The factors chosen for this run
satisfy the best values suggested by the main effects shown
in Table 4, i.e. maximization of the cheek thickness, the
cooling to the disc surfaces and the conductivity; therefore
a confirmation run was not deemed necessary. Moreover,
the vent size was minimized during run 6, thus creating a
small bridge gap (and hence good caliper stiffness). A
brake disc manufactured to the design and material as laid
down in run 6 would give a mass saving of about 52 per
cent over a conventional cast iron disc. However, it would
be extremely difficult to manufacture a disc with 1.0 mm
vents owing to casting constraints. A more realistic vent
size of 3.0 mm is suggested which would give a mass
saving of about 42 per cent. This would increase the overall
disc thickness to 27 mm which could be accommodated by
increasing the present caliper bridge gap by 4.3 mm.

7.0
7.0
7.0
7.0
12.0
12.0
12.0
12.0

Run

1
2
3
4
5
6
7
8

1.0
1.0
11.0
11.0
1.0
1.0
11.0
11.0

A,
cheek thickness
(mm)

B,
vent width
(mm)

Table 3 Results of L8 Taguchi study of an Al-MMC brake disc

Mean,
maximum
temperature
(8C)

Variance

S=N

D G GRIEVE, D C BARTON, D A CROLLA AND J T BUCKINGHAM

Mass
saving
(%)

250

PARAMETRIC STUDY OF Al-MMC DISC


DESIGN

Of the three critical factors highlighted in the Taguchi


study, the disc cheek thickness was identified as having the
greatest effect on the peak temperature and is also the
easiest factor to change in practice. In contrast, increasing
the cooling air speed would require special ducting to the
brake disc and redesign of the vehicle's front end. The
conductivity is linked to the composition of the chosen disc
material and so cannot be changed significantly without
altering other material properties. The cheek thickness can
be varied early in the design process with the only
restrictions being the mass saving that is achieved and the
packaging and size constraints. Parameter studies were
therefore carried out in which the cheek thickness and the
vent width were varied simultaneously so as to limit the
total disc thickness (caliper bridge gap) to 27 mm. This was
considered to be the largest gap that could realistically be

Proc Instn Mech Engrs Vol 212 Part D

D01097 # IMechE 1998


Downloaded from pid.sagepub.com at CIDADE UNIVERSITARIA on March 27, 2015

DESIGN OF A LIGHTWEIGHT AUTOMOTIVE BRAKE DISC USING FE AND TAGUCHI TECHNIQUES

Table 4

251

Responses and main effects for the signal-to-noise ratios


Value for the following factors

Level
1
2
Main effects

A,
cheek thickness
55.4
52.9
2.5

B,
vent width
54.6
53.8
0.8

accommodated without undue loss of caliper stiffness or


increase in caliper mass. Again simulations of the Alpine
descent and autobahn stop were used to assess the effect of
cheek thickness for a number of proprietary Al-MMC
materials assuming a typical GVM of 1830 kg.
Increasing the disc cheek thickness from 8 to 12 mm was
found to reduce peak Alpine descent temperatures by about
16 8C for all materials considered (Fig. 4). However, only
the Al-MMC materials containing very-high-percentage
SiC reinforcements or high-temperature matrix alloys, 68%
SiC Al-MMC and 7% TiB2 Al-MMC respectively, were
predicted to achieve peak temperatures below their MOT
for all dimensional combinations with finite vents. The
large proportion of ceramic present in the 68% Al-MMC is
believed to improve the high-temperature resistance of the
composite by increasing its softening temperature and
hence its MOT. The 7% TiB2 Al-MMC was believed to be

C,
density
54.5
53.8
0.7

D,
conductivity

E,
specific heat

54.6
53.7
0.9

54.4
54.0
0.4

F,
cooling rate
54.7
53.6
1.1

successful because the inert nature of the TiB2 particles


allows low-silicon alloys to be specified with better hightemperature properties. The higher conductivity of the
aluminium-based composites is thought to reduce the
temperatures seen at the disc surfaces by rapidly conducting heat to other parts of the brake geometry. This theory is
supported by examining Table 5, which indicates that
materials with the higher conductivities produce lower
temperatures.
The results for the autobahn stop (Fig. 5), predict that all
the materials studied are capable of surviving the test if a
suitable disc geometry is chosen; because of the dependence of this test upon the thermal capacity of the disc,
thicker discs are simply more capable of absorbing the
thermal energy generated during high-speed braking.
Even with overall disc thicknesses of 27 mm, Table 6
shows that significant mass savings can be achieved with

Fig. 4 Predicted peak temperatures for vented brake discs 27 mm wide during the Alpine descent constructed
from a selection of brake disc materials. The conductivities, k, are shown for each material
D01097 # IMechE 1998

Proc Instn Mech Engrs Vol 212 Part D


Downloaded from pid.sagepub.com at CIDADE UNIVERSITARIA on March 27, 2015

132
86.2
101
98.6
125
195

265

206
338
215
310
232/379 (T6)
325
228
225

303
215
296
120/274 (T6)
150
.210
150

97.8
151
175
126.4
156
160
220
170
218
828
879
837
838
820
899
810
722
882
538
454
450
454
480
510
810
520
500
3030
2711
2760
2770
2780
2950
3040
3000
2818
Infiltration of ceramic pre-form
Sand cast
Sand cast
Sand cast
sand cast
Infiltration of ceramic pre-form
High-pressure infiltration
Infiltration of ceramic pre-form
Sand cast with salt refiner

MOT
(8C)
Density
(kg=m3 )
Manufacturing
route
Material

all the MMC materials studied. Brake discs manufactured


from the 68% SiC Al-MMC or 7% TiB2 Al-MMC to either
the 12312 or 11511 (vaneventvane) geometries
would give mass savings of 44.441.4 and 48.645.7 per
cent respectively, which for a typical cast iron disc having a
mass of about 5 kg would give a saving of between 2.07
and 2.43 kg per disc.
Eliminating the vents altogether in the 13.5013.5
combination enabled an assessment of their contribution to
the overall cooling to be made. In the Alpine descent, the
disc brake temperatures were increased by about 35 8C
because of the loss of cooling from the vent surfaces. This
study confirms that vents play an important role in
temperature control during the Alpine descent and should
be included in any design even if only at a minimal width
(3 mm) so as to allow a nominal air flow. In the autobahn
stop, removing the vents had the opposite effect of reducing
the disc brake temperatures owing to the increased thermal
mass created by the extra cheek thickness. It must be noted,
however, that the study only examined a single autobahn
stop. Multiple stops would test the disc's ability to recover
thermally between stops and cooling vents would no doubt
be important for this recovery process.

30 vol %, Al2 O3, Al1% Mg matrix, 93-X-3050


10 vol %, SiC, A359 matrix, F3S.10S T61
15 vol %, 20 m SiC Al9% Si3 Cu matrix
20 vol %, SiC, A359 matrix, F3S.20S-T6
30 vol % SiC, Al10% Si1% Mg matrix, 94-X-4040-30P-T6
55 vol % SiC, AlSiMg matrix, 92-X-2039
68 vol % -SiC, 2014 matrix alloy, HIVOL B
70 vol % SiC, AlSiMg matrix, 90-X-027
Al 6101 matrix alloy 7 vol % TiB2

Modulus
(GPa)
Ultimate
tensile
strength
(MPa)
Yield
strength
(MPa)
Specific
heat
capacity Conductivity
(J=Kg K)
(W=m K)

Table 5 Material properties of currently available Al-MMCs suitable for a brake rotor application

15.1
20.7
18.8
17.5
14.9
10.0
6.9
6.2
23.5

D G GRIEVE, D C BARTON, D A CROLLA AND J T BUCKINGHAM


Coefficient
of thermal
expansion
(10 6 K)

252

CONCLUSIONS

FEA studies of peak braking temperatures for Al-MMC


discs fitted to a medium-sized passenger vehicle have
predicted that a GVM of less than 1100 kg is required to
enable the surface of the disc to remain undamaged during
certain critical brake tests. However, no modifications were
made at this stage to the current cast iron disc design to
take advantage of the MMC's properties. By combining a
FEA with a Taguchi analysis, a technique has been demonstrated which quickly and cost-effectively enables a number
of design and material factors to be considered in a search
for an optimized Al-MMC brake disc. The Taguchi analysis
revealed that the following factors are the most influential
in order of importance: the disc cheek thickness, the
cooling rate to the disc surfaces and the conductivity of the
disc material.
Parameter sensitivity studies of the effect of cheek
thickness and vent width indicated that the 12312 mm
(cheekventcheek thickness) design produced the lowest
temperatures for both Alpine and autobahn tests. Maximization of the cheek thickness is beneficial for the autobahn
stop but vents were shown to be critical to the thermal
performance of Al-MMC rotors, their surface area providing vital cooling during long stops such as the Alpine
descent. The minimum vent width is ultimately limited by
the core strength used to produce the vents during the
casting process and it is therefore recommended that vents
be retained at a nominal width of 3 mm.
The same parameter studies highlighted two types of AlMMC material that would pass the legally required tests

Proc Instn Mech Engrs Vol 212 Part D

D01097 # IMechE 1998


Downloaded from pid.sagepub.com at CIDADE UNIVERSITARIA on March 27, 2015

DESIGN OF A LIGHTWEIGHT AUTOMOTIVE BRAKE DISC USING FE AND TAGUCHI TECHNIQUES

253

Fig. 5 Predicted peak temperatures for vented brake discs 27 mm wide during the autobahn stop constructed from
a selection of brake disc materials. The thermal capacities, rC p , are shown for each material

Table 6 Disc masses and percentage mass savings over present cast iron design
8 mm cheeks,
11 mm vents

Material

Disc
mass
(kg)

Cast iron
5.81
2.43
30% Al2 O3 Al-MMC
10% SiC Al-MMC
2.17
15% SiC Al-MMC
2.19
20% SiC Al-MMC
2.22
30% SiC Al-MMC
2.23
68% SiC Al-MMC
2.44
2.26
7% TiB2 Al-MMC
 Present cast iron disc mass 5:25 kg.

9 mm cheeks,
9 mm vents

10 mm cheeks,
7 mm vents

11 mm cheeks,
5 mm vents

12 mm cheeks,
3 mm vents

13.5 mm cheeks,
0 mm vents

Mass
savings
(%)

Disc
mass
(kg)

Mass
savings
(%)

Disc
mass
(kg)

Mass
savings
(%)

Disc
mass
(kg)

Mass
savings
(%)

Disc
mass
(kg)

Mass
savings
(%)

Disc
mass
(kg)

Mass
savings
(%)

10.7
53.7
58.7
58.3
57.7
57.5
53.5
57.0

6.19
2.59
2.31
2.33
2.37
2.37
2.60
2.41

17.9
50.7
56.0
55.6
54.9
54.9
50.5
54.1

6.57
2.75
2.46
2.47
2.51
2.52
2.76
2.55

25.1
47.6
53.1
53.0
52.2
52.0
47.4
51.4

6.95
2.91
2.60
2.62
2.66
2.67
2.92
2.70

32.4
44.6
50.5
50.1
49.3
49.1
44.4
48.6

7.34
3.07
2.74
2.76
2.80
2.81
3.08
2.85

39.7
41.6
47.7
47.4
46.6
46.4
41.4
45.7

7.91
3.31
2.96
2.98
3.02
3.03
3.32
3.07

50.7
37.0
43.7
43.3
42.4
42.2
36.8
41.4

with appropriate design modifications to the disc: composites with a high level of reinforcement, e.g. 68% SiC in an
Si-based matrix alloy, and composites using a temperatureresistant matrix alloy (due to a low Si content), e.g. 7%
TiB2 -reinforced Al-MMC.

technical support and advice. The Department of Trade and


Industry LINK Structural Composites Programme and the
Engineering and Physical Sciences Research Council are
also thanked for supporting this research project.

REFERENCES

ACKNOWLEDGEMENTS
The authors would like to thank Mr S. Boulton and his
research staff at BBA Friction, Cleckheaton, for their

1 Neitzl, B., Barth, M. and Matic, M. Weight reduction of disc


brake systems with the utilisation of new aluminium material.
SAE technical paper 940335, 1994, pp. 2935.

D01097 # IMechE 1998

Proc Instn Mech Engrs Vol 212 Part D


Downloaded from pid.sagepub.com at CIDADE UNIVERSITARIA on March 27, 2015

254

D G GRIEVE, D C BARTON, D A CROLLA AND J T BUCKINGHAM

2 De Sanctis, A. M., Evangelista, E., Forcellese, A. and


Fuganti, A. Forging of a MMC for an automotive component.
In Proceedings of the Symposium on Materials for Lean Weight
Vehicles, University of Warwick, 1995, pp. 7382 (Institute of
Materials, London).
3 Wycliffe, P. Friction and wear of Duralcan reinforced aluminium composites in automotive braking systems. SAE technical
paper 930187, 1993.
4 Dwivedi, R. Performance of MMC rotors in dynamometer
testing. SAE technical paper 940848, 1994, pp. 6571.
5 Grieve, D. G., Barton, D. C., Crolla, D. A., Buckingham, J.
T. and Chapman, J. Investigation of light weight materials for
brake rotor applications. In Proceedings of the Symposium on
Materials for Lean Weight Vehicles, University of Warwick,
1995, pp. 6372 (Institute of Materials, London).

6 Grieve, D. G., Barton, D. C., Crolla, D. A., Chapman, J. and


Buckingham, J. T. Light weight disc brake materials. In
Advances in Automotive Braking Technology, 1996, pp. 89105
(Mechanical Engineering Publications, London).
7 Dennis, R. W., Newstead, C. and Ede, A. J. The heat transfer
from a rotating disc in an air crossflow. In Proceedings of the
Fourth International Heat Transfer Conference, Paris, 1970, Vol.
8, 1970, p. 134 (Elsevier, Amsterdam).
8 Bailey, T. P., Buckingham, J. T. and D'Cruz, A. H. Optimisation of brake disc design using thermal imaging and finite
element techniques. In Autotech 1991, Birmingham, 1991
(Mechanical Engineering Publications, London).
9 Taguchi, G. Taguchi on Robust Technology Development, 1993
(ASME Press, New York).

Proc Instn Mech Engrs Vol 212 Part D

D01097 # IMechE 1998


Downloaded from pid.sagepub.com at CIDADE UNIVERSITARIA on March 27, 2015